Estimating the Lightning Performance of a Multi- Circuit Transmission Tower

Transcription

1 Estimating the Lightning Performance of a Multi Circuit Transmission Tower Pawel Malicki, Andrzej Mackow and Mustafa Kizilcay University of Siegen Chair of Electrical Power Systems Siegen, Germany pawel.malicki@unisiegen.de Abstract Lightning strokes cause mainly outages of transmission s. A lightning study takes into consideration possible impacts of lightning strokes on transmission systems. The lightning performance of multicircuit transmission with AC and DC circuits is investigated. Till today lightning analysis of HVDC and HVAC systems were done separately. Since both systems will be installed on the same, lightning performance analysis of multicircuit that has been converted into a hybrid has become very challenging. Present paper illustrates how to combine efficiently different methods to support insulation coordination studies. Lightning incidence and range of outages for two configurations of hybrid have been estimated using electrogeometric and simulations in EMTPATP. Keywords lightning strokes, flashover, lightning attachment, shielding, ling I. INTRODUCTION An interest in lightning research has increased in last years. The reasons have different origin. Firstly more field measurements of lightning strokes are available nowadays. Worldwide wellestablished distributions of lightning current parameters are based on measurements in years by research group in Switzerland [1]. Since then new measurements of lighting strokes have been conducted in Japan [2], Brazil [3]. Measuring equipment has become more advanced than 30 years ago. Thus quality of measurements of lightning parameters is more reliable. Some of measurements agree coincide with measurements in Switzerland [2], whereas other deviate considerably from them [3]. New design of power transmission s is another factor that makes lightning studies still necessary. Lightning performance of pylon transmission s [4] and hybrid HVAC/HVDC [5] s has to be estimated before they will be erected. Particularly the hybrid with AC and DC systems offers an interesting alternative that includes two different types of power transmission [6]. The importance of insulation coordination studies of a hybrid was noted in [5]. Conversion of one system into new HVDC system along an existing route on the same will be taken into consideration. Available conductors, shield wires and insulators strings of AC s will be adapted to a HVDC system. The layout of the led s A is shown in Fig. 1. That may have one or two shield wires. Comparison of shielding performance against lightning with one or two shield wires is investigated in this paper. Transmission with two shield wires presented in Fig. 1 is named A'. The upper left crossarm carries a bipolar 420kV HVDC system. Evaluation of the risk of lightning stroke outage of transmission s is very important. The outages that are caused by lightning stroke have two origins. Lightning strokes that are intercepted by shield wires can cause backflashover across insulator strings. Due to shielding failure a lightning stroke may hit a phase conductor and can cause a flashover across an insulator. Shielding failures that are estimated by electrogeometric s are investigated. This study employs general expression for the estimation of lightning incidence, maximum shielding current and lightning outage rates for the HVDC/HVAC hybrid. The transients program EMTPATP [7] is well suited to analyze lightning surge phenomenon on overhead s. N SW W TOWER A V U SW N SW TOWER A Figure 1. Tower layout of investigated hybrid A and A' W V U

2 II. LIGHTNING OCCURENCE A. Types of Lightning Discharges About threequarters of lightning flashes do not hit the ground. These are termed cloud flashes. Lightning discharges between cloud and earth are termed cloudtoground discharges. Four types of lightning discharges between cloud to earth and earth to cloud are shown in Fig. 2. Only the initial leader is shown for each type. About 90 % of global cloudtoground lightning are downward negative lightning flashes and remaining 10 % of global cloudtoground lightning are downward positive lightning flashes according to [13]. Downward flashes exhibit downward branching, while upward flashes are branched upward. Thus only negative downward lightning is considered in this study. a) b) c) d) Figure 2. Types of lightning discharges a) Downward negative lightning; b) Downward positive lightning; c) Upward positive lightning; d) Upward negative lightning B. Flash Density In the present study the ground flash density N g (strikes/km 2 /year) has been estimated from records of lightning locating system BLIDS [12] in Germany. This lightning locating system is based on the timeofarrival method (TOA). TOA uses the measurements of the timeofarrival of electromagnetic field at several stations. A lightning stroke generates an electromagnetic field which propagates with the speed of light. Comparisons of the differences in the arrival time of two or more stations define the source location. Therefore stations must be precisely synchronized. Other lightning locating systems work with magnetic directionfinders (MDF) or with the combination of TOA and MDF [22]. If no measurements of the ground flash density N g are available, this parameter can be roughly estimated from the annual number of thunderstorm days, also called the keraunic level. C. Lightning Analysis The functionality of shield wire can be estimated with Shielding Failure Rate (SFR). SFR is number of lightning strokes (strikes/100 km/year) that can directly terminate on the phase conductor. In other words it indicates malfunctions of shield wire. SFR = 2N g L I max 3kA D C (I)f(I)dI 1 where: N g is ground flash density (strikes/km 2 /year) according to [12], L is length (km), I max is the maximum current, that can be estimated with electrogeometric, D C(I) is shielding failure width, f(i) is the probability density function of the lightning crest distribution. Not all of these lightning strikes to phase conductor would result in flashover across insulators on a crossarm. The number of lightning strokes to phase conductor that cause flashover across insulator is called Shielding Failure Flashover Rate (flashovers/100 km/year) and can be calculated with (2). I max SFFOR = 2N g L D C (I)f(I)dI (2) I C where: N g is ground flash density (strikes/km 2 /year) according to [12], L is length (km), I max is the maximum current, that can be estimated with electrogeometric, I C is shielding critical current (ka), D C(I) is shielding failure width, f(i) is the probability density function of the lightning crest distribution. Lightning current parameters and probability density function of the lightning crest distribution are based on the lightning current distribution according to [11]. In Table I different probability density functions for negative first strokes according to [13] are presented. TABLE I DIFFERENT PROBABILITY DENSITY FUNCTIONS Probability density function f(i) [13] Location Median (ka) σlgi Berger Switzerland Takami Japan Visacro Brazil Global 7 countries where: σlgi is standard deviation of probability density function The choice of the probability density function f(i) has significant influence to SFR and SFFOR. The global distribution of first negative strokes is furthermore recommended in [13] and is considered in this investigation. III. LIGHTNING ATTACHMENT MODEL The maximum shielding current I max can be estimated with different lightning attachment s like electrogeometric, Eriksson s, generic and statistical according to [23]. In this work maximum shielding current I max was estimated on the basis of the electrogeometric (EGM) for different transmission geometry. The effect of the struck object height on striking distance is neglected in the

3 Height of Tower (m) Lightning current (ka) electrogeometric. General concept of electrogeometric was presented in [14]. In Fig. 3 radii r c are drawn from shield wire and phase conductor for increasing lightning currents. Additionally a horizontal a distance r g is drawn from earth surface. Those radii are striking distances for a vertically moving lightning stroke and are dependent on the stroke current crest value. The intersections are marked A, B and C. The distances D C and D G are the exposure distance for the phase conductors and shield wires. For increasing lightning C D G The lowest value of current I min in (1) is 3 ka according to CIGRE data [11]. Various values of coefficients A and b have been proposed by researches over years to calculate r c and r g. The basic formulas for striking distances are [8]: r c = A C I b c (3) r g = A g I b g (4) Lightning incidence is the next quantity to determine for the lightning performance of a transmission. The annual number of lightning strikes to shield wires per 100 km of a transmission can be given as a simplified expression D G D C=0 N s = 0.1 N g (2 R eq b) 5 C C DG B DC D C B A B A where: N g (strikes/km 2 /year) is the ground flash density according to [12], b (m) is the separation distance between shield wires and R eq (m) is the equivalent interception radius of the shield wire. The equivalent interception radius of a conductor can be defined as rc R eq = r h E 6 h β y rc A rg r g r g where: h (m) is the height of the conductor with the coefficients r and E. The basic formulas for calculating the factors r and E are given in [14]. In this paper coefficients for EGM and R eq in Table II are used [15]. Distance (m) TABLE II. COEFFICIENTS FOR EGM Figure 3. Sketch of striking distances currents the distance D C decreases until a point is reached at which D C becomes zero. This point is defined by the current I max. In Fig. 4 a downward leader in the shaded area will terminate in the phase conductors. According to the lightning attachment s only one phase can be hit directly by the lightning stroke. Consequently only at that phase a flashover may occur. EGM rc (m) rg (m) Req (m) Young 29 I I h 0.44 Love 10 I I h 0.46 Armstrong, Whitehead 6.72 I I h 0.29 Brown, Whitehead 7.1 I I h 0.29 Wagner, Hileman 14.2 I I h 0.43 Whitehead 9.4 I I h 0.46 β Distance (m) Figure 4. Interception zones of the electrogeometric Imax Imin IV. SIMULATION MODEL The simulation for the lightning analysis is based mainly on a proposed in [16]. That was developed based on several field measurements and tests. Moreover it allows considering most relevant factors that influence propagation of lightning surge on overhead and transmission. The simulation shown in Fig. 5 enables estimation of shielding critical current. All overhead s on the same are represented by the Constant Parameter Distributed Line (CPDL) at f = 400 khz. This frequency is in the range of main resonant frequency of travelling waves in a span of length 330 m. To investigate lightning performance of the hybrid for two different layouts (Fig. 1), a section with 9 s is selected in order to take into consideration of reflected/refracted waves from adjacent s realistically.

4 7km CPDL 13phase (330m) 13phase l (130m) 13phase (200m) 13phase (330m) 7km CPDL 2 x shield wire Flashover I(t) Zf=1000Ω shield wire 2 x No. 1 No. 5 lightning current No. 6 No. 9 Figure 5. Simulation for determination of the critical current due to direct lightning stroke It is assumed that a lightning stroke hits this section between 5 and 6. At both ends of this led section a section of 7 km with the same electric parameters has been added to delay additionally traveling waves reflected at the voltage sources (Fig. 5). So that they do not influence the results in the time window of analysis. The lightning stroke is led by a current source and a parallel resistance of 1000 Ω, which represents the lightning path surge impedance according to [16]. Stroke to a span is more likely to occur than a stroke to a according to lightning observation in [18]. Procedure to estimate the critical current was proposed in [14]. Simulation in EMTPATP allows using that procedure to obtain the critical current for outermost conductors of 380kV circuit. Moreover that method considers effects like footing resistance, coupling from lightning current that flows through struck shield wire, dependencies of front time and maximal steepness related to crest current value. The footing impedance was assumed to 10 Ω according to [16]. The power frequency voltage is considered by calculating the critical current for struck phase for instantaneous power frequency voltages estimated for each of twelve 30 steps of phase angle. Thus sinusoidal waveform of voltage is considered. Mean value of twelve estimated critical currents is inserted in (2) as lower integration limit. Leader progression (LPM) is appropriate way to represent flashover across insulator. LPM considers different phases of flashover phenomenon (streamer and leader phases). After streamer has bridged a gap leader progression starts. That transition period can be estimated with leader onset condition (s. TABLE III). Moreover equations to calculate leader velocity and leader length are provided in TABLE III. TABLE III. Leader onset condition Leader velocity Leader length E 0p LEADER PROGRESSION CONDITIONS Leader Progression Model [19] u(t) E 0 D v l = 170 D ( u(t) D l l E 0p ) e ( u(t)/d) l l = v l (t)dt 670 kv/m Gap length D of insulator strings for 110kV, and 420kV HVDC is about 1000 mm and 3000 mm, respectively. The layouts of the ed s A and A' are shown in Fig. 1. The sections were represented by lossless Constant Parameter Distributed Line (CPDL) [9]. In Fig. 6 the used Multistory [21] is shown to represent transmission s. Figure 6. Multistory h2 h1 r3 r1 r2 In multistory each vertical section between cross arms is represented by a lossless connected in series with RL parallel circuit. This parallel circuit represents attenuation of traveling waves. Calculation of surge impedance of the Z T1Z T4 with formula R1 R3 R4 Z Ti = 60 ln [cot {0.5 tan 1 ( R )}] (7) h The RL values are determined as functions of surge impedance [21]. CIGRE waveform of concave shape has variable front time T d30, constant time to half value T h = 77.5 μs and variable steepness S m [13]. According to [11] the maximum steepness (8) and the front time (9) depend on the peak value of the lightning current. ZE S m = 3.9 I 0.55 (8) T d30 = I (9) L1 L2 L3 L4 x1 x2 x3 x4

5 Lightning current I max (ka) Voltage (kv) In Fig. 7 three different lightning current waveforms with variable steepness and front time are shown. Lightning Current (ka) (1) (2) Time (μs) Figure 7. Lightning Current (1) 12 ka; (2) 14 ka ; (3) 18 ka V. RESULTS Lightning incidence N S to shield wire(s) is calculated for various EGM given in Table II. The results are shown in Table IV. N g is based on [12] and amounts to 4 strikes/km 2 /year. Results are similar among all interception s and are higher for A' with double shield wires due to higher interception capability. TABLE IV. (3) LIGHTNING INCIDENCE RESULTS Req (m) Ns (strikes/100 km/year) Tower A Tower A' Young 14.3 h Love 13.9 h Armstrong, Whitehead 33.7 h Brown, Whitehead 30.1 h Wagner, Hileman 11.0 h Whitehead 14.2 h The maximal current as crest value that can directly strike phase conductor is determined by means of electrogeometric s. In Fig. 8 maximum shielding currents among various EGM are summarized. The variation of maximum shielding currents among lightning attachment s is considerably large. Tower A' is equipped with double shield wire and intercepts efficiently lightning strokes. Thus maximal lightning current for A' is considerably lower among all EGMs. Since EGM considerably influences SFR and SFFOR performance of the based on parameter I max, several EGM have been taken into account to show a range of SFR and SFFOR values depending on EGM Young Love Armstrong, Whitehead Tower A' Brown, Whitehead Tower A Wagner, Hileman Whitehead Figure 8. Maximum crest value of the current of a direct lightning stroke to uppermost phase conductor. The calculation of the critical current for AC circuits is more complex, since power frequency voltage is timevarying. The power frequency voltage is considered by calculating the critical current for struck phase for instantaneous power frequency voltages estimated for each of twelve 30 steps of phase angle (s. Fig. 9). The critical current of each phase depends mainly on the phase angle of power frequency voltage at the stroke instant. In Fig. 9 the value of twelve estimated critical currents is shown. The mean value is inserted in (2) as lower integration limit. In case of HVDC circuit poletoground crest voltage is considered. 326 kv 326 kv critical current IC 11 ka critical current IC 12 ka ωt Figure 9. Critical current (380 kv AC System) for each of twelve 30 steps of phase angle Calculation of critical current for HVDC systems requires only one computation, since voltage of positive pole is constant and equal to 420kV. Critical current was obtained by a simulation and amounts to ka for 380kV and is about 14 ka for HVDC. In Fig. 10 (1) voltage difference across the insulator string of plus pole of HVDC causing flashover after shielding failure is presented. Voltage Across Insulator (MV) (1) (2) Time (μs) Figure 10. Voltage difference across the insulator string of plus pole causing (1) flashover and (2) no flashover due to shielding failure Maximum shielding currents that were estimated with EGM are employed for shielding performance calculation. SFR and SFFOR are presented in Table V and Table VI. SFR is the same for both sides of despite different systems. Converting of a HVAC system into a HVDC system does not change the location of conductor or length of insulators. The length of highest crossarms for each is the same and this is valid for both sides of the. Various values of maximal shielding current I max and shielding width D C(I) influence shielding performance of the investigated s.

6 SFR values estimated with different EGM vary from each other significantly. Two shield wires in A' reduce efficiently number of strokes that may directly terminate in outermost phase conductors. SFFOR is different for two sides of the investigated s. Following a direct stroke in upper outmost conductor flashover across insulator to crossarm at AC system is more likely. SFFOor A exceeds recommended SFFOor s serving critical loads [14]. It is recommended SFFOR of 0.05 flashover/100 km/year. In particular, A with one shield wire does not fulfil this requirement. TABLE V. SFR (STRIKES/100 KM/YEAR) AND SFFOR (FLASHOVER/100 KM/YEAR) FOR TOWER A Tower A SFR SFFORHVDC SFFOR380kV Young Love Armstrong, Whitehead Brown, Whitehead Wagner, Hileman Whitehead TABLE VI. SFR (STRIKES/100 KM/YEAR) AND SFFOR (FLASHOVER/100 KM/YEAR) FOR TOWER A Tower A' SFR SFFORHVDC SFFOR380kV Young Love Armstrong, Whitehead Brown, Whitehead Wagner, Hileman Whitehead VI. CONCLUSION Shielding performance calculations have been performed for two s coming into consideration for a multicircuit with HVAC/HVDC circuits. Several lightning interception and attachment s were implemented to evaluate SFR and SFFOR. The maximum shielding current depends strongly on geometry and lightning attachment used. Results differ considerably depending on the electrogeometric used. Shielding performance of a can be efficiently increased by additional shield wire. Tower A' with two shield wires offers substantially low values of SFR and SFFOR. Conversion of an existing with only HVAC systems into HVAC/HVDC multicircuit decreases probability of occurrence of flashover due to direct lightning stroke to outermost conductor. Whereas SFR after conversion into HVDC remains unchanged, SFFOR is slightly reduced for the HVDC system. Constant 420 kv voltage of plus pole makes upper insulator at plus pole less prone to flashover by a direct stroke to phase conductor in comparison with phase conductor of original 380 kv system. REFERENCES [1] K. Berger, R. B. Anderson und H. Kroninger, Parameters of lightning flashes, Electra, Bd. 80, pp. 2337, [2] J. Takami und S. Okabe, Observational results of lightning current on transmission s, IEEE Trans. Power Del., Bd. 22, pp , [3] S. Visacro, C. R. Mesquita, A. De Conti und F. H. Silveira, Updated Statistics of lightning currents measured at Morro de Cachimbo station, Atmos. Res., Bd. 117, pp. 5563, [4] J. Tohid, B. Leth, S. Filipe Miguel Faria da, E. Brian, Holbøll, Joachim, Assessment of Lightning Shielding Performance of a 400 kv Double Circuit Fully Composite Pylon, Cigré 2016 Paris Session, 2016, (submitted) [5] CIGRE WG B2.41, Guide to the conversion of existing AC s to DC operation, [6] CIGRE JWG B2/C1.9, Increasing Capacity of Overhead Transmission Lines: Needs and Solutions, [7] Canadian/American EMTP User group, ATP Rule Book, Portland, USA: revised and distributed by EEUG Association, [8] IEEE WG on Estimating the Lightning Performance of OHTL, IEEE Guide for Improving the Lightning performance of Transmission Lines, IEEE, New York, [9] Canadian/American EMTP User group, ATP Rule Book, Portland, USA: revised and distributed by EEUG Association, [10] CIGRE WG C4.501, Guide for numerical electromagnetic analysis methods: Application to surge phenomena and comparision with circuit theorybased approach, CIGRE, [11] CIGRE WG 01 SC 33, Guide to procedures for estimating the lightning performance of transmission s, [12] BLIDS: Flash Information Service by SIEMENS [On]. [13] CIGRE WG C4.407, Lightning Parameters for Engineering Applications, CIGRE, [14] A. R. Hileman, Insulation Coordination for Power Systems, Boca Raton, USA: CRS Press, [15] P. N. Mikropoulos and T. E. Tsovilis, Estimation of lightning incidence to overhead transmission s, IEEE Transactions on Power Delivery, pp , July [16] A. Ametani and T. Kawamura, A method of a lightning surge analysis recommended in Japan, IEEE Transactions on Power Delivery, pp , [17] A. Mackow, M. Nilges, M. Kizilcay and D. Potkrajac, Analysis of Backflashover across Insulator Strings of a Multicircuit Transmission Tower with AC and DC systems, in Proceeding of EEUG Conference 2014, Cagliari, Italy, [18] J. Takami and S. Okabe, Observational results of lightning current on transmission s, IEEE Transaction on Power Delivery, pp , [19] A. Pigini, G. Rizzi, E. Garbagnati, A. Porrino and G. Pesavento, Performance of large air gaps under lightning overvoltages: Experimental study and analysis of accuracy of predetermination methods, IEEE Transations on Power Delivery, pp , April [20] L. Dube, Users guide to MODELS in ATP, [21] M. Ishi, T. Kawamura, T. Kouno, T. Ohsaki, E. Shiokawa and K. Murotani, Multistory transmission for lightning surge analysis, IEEE Transactions on Power Delivery, pp , [22] CIGRE WG C4.404, Cloud to ground lightning parameters derived from lightning location systems, CIGRE, [23] P. N. Mikropoulos and T. E. Tsovilis, Lightning Attachment Models and Maximum Shielding Failure Current: Application to Transmission Lines, PowerTech, Bucharest, June 2009.